Fourier Approximation of magMEMS Oscillations: Neural Network Space Handling

The initial value problem for a model of magnetic Micro-Electro-Mechanical System (magMEMS) with current-carrying conductors is considered. The model equation is a conservative MEMS oscillator due to the presence of the singular term corresponding to the magnetic force between two current-conducting wires. If the excitation parameter is below a certain threshold, the actuator oscillates. Fourier approximation of periodic solutions is enhanced by neural network space handling for both the limiting case where the geometric parameter approaches zero and the general case with arbitrary geometric parameter values. The method is validated through numerical comparisons with high-precision solutions, and its limitations are identified during the verification of experimental results. The findings can be useful for the design of magMEMS models.